presentación de powerpointbijker/franco70/talks/wolf.pdf · hic vivunt leones!). conserved....

Post on 19-Jun-2020

4 Views

Category:

Documents

0 Downloads

Preview:

Click to see full reader

TRANSCRIPT

LINEAR CANONICAL TRANSFORMS:LINEAR CANONICAL TRANSFORMS:40 years of developments40 years of developments

Kurt Bernardo WolfInstituto de Ciencias Físicas

Universidad Nacional Autónoma de México CuernavacaCuernavaca

Beautyin Physics

Cocoyoc, 14—18 May 2012

Back in 1970…the same formulas were found at the same time in two very different places:

Stuart A. Collins Jr. Stuart A. Collins Jr. (Electroscience Lab, Ohio State University, Columbus)

described light propagation in the paraxial régime through an integral kernel, built from thin lenses and empty spaces.Lens-system diffraction integral written in terms of matrix opticsJ. Opt. Soc. Am. 60, 1168—1177 (1970)

Marcos Moshinsky Marcos Moshinsky and

Christiane Quesne Christiane Quesne (Instituto de Física, UNAM)

sought for the conservation of uncertainty under linear transformations of phase space, as a matter of intrinsicmathematical interest.Oscillator Systems. In: Proceedings of the 15th Solvay Conference in Physics (1970)Linear canonical transformations and their unitary representationJ. Math. Phys. 12, 1772-1780, 1780—1783 (1971)

Stuart A. Collins Jr.Professor Emeritus

>514 citations (Feb 2012)

The ElectroScience Laboratory

The Ohio State University

Columbus, Ohio

OptoelectronicsSpace Science

6 patents from 1982 to 2008

Lens-System Diffraction Integral Written in Terms of Matrix OpticsJ. Opt. Soc. Am. 60, 1668-1177,1970

1st

and 2nd

–order citations to the paper of Stuart Collins

citationsper year

Marcos MoshinskyInstituto de Física

Universidad Nacional Autónoma de México

Christiane QuesneFaculté

de SciencesUniversité

Libre de Bruxelles

Oscillator Systems. In: Proceedings of the 15th Solvay Conference in Physics (1970)Linear canonical transformations and their unitary representationJ. Math. Phys. 12, 1772-1780, 1780—1783 (1971)>346 citations (Feb 2012)

1st

and 2nd

–order citations to the paper of Moshinsky & Quesne

citationsper year

The paraxial  optical construction with matrices(Brower 1964, Gerrard & Burch 1975)

}The Fresnel transformis integral, ergo:

1.

Find the excess quadratic phase of off‐axis rays in compound systems.2.

Find the Lagrangian

that requires

3.

Find the phase

(only

from the Fresnel transform). 4.

Find normalization

from energy conservation.

x:  ray position,  p:  n

x

ray slope

Lens-system diffraction integral written in terms of matrix opticsJ. Opt. Soc. Am. 60, 1168—1177 (1970)

The conservation of uncertainty

under linear transformations  of  w = ( ),xp

Preservation of HW  =>symplectic conditions.Integral transform formacting on  xf and on pf =>2N

differential equations

for the kernel:for the kernel:

Limit to the identity implies

Oscillator Systems. In: Proceedings of the 15th Solvay Conference in Physics (1970)Linear canonical transformations and their unitary representationJ. Math. Phys. 12, 1772-1780, 1780—1783 (1971)

The well‐known 1‐dim LCTs

…be careful with phases:

LCTs in the Hilbert space L (R)are unitary,

They compose as Sp(2,R)‐‐up to a metaplectic sign

They include the Fourier

transform‐‐but for a phase, so they coverthe Fourier cycle twice.

2

The metaplecticmetaplectic

phasehas bedevilled countless paraxial optical papers…because their authors did not realize that  Sp(2N,R)has an infiniteinfinite

cover group, of which the integral realization is the double

cover.

Valentin Bargmannhas shown thatif we use the polarmatrix decompositionwe can parametrizeany cover group (1947)

…actually, this was realized several years later…We were concerned with the harmonic oscillatorrather than with the fractional Fourier transform,and optics people do not care about overall phases.

The Lie algebra of the group of integral transformshas as generators second‐order differential operators

The

Lie algebra thatgenerates

LCTs:

Is a realization of sp(2,R)associated withparaxial optical elements

There exist Bargmann‐typeHilbert spaces ofanalytic functionswhere Complex LCT’sare unitary.

KBW, Canonical transforms I. Complex linear transforms, J. Math. Phys. 15, 1295-1301 (1974)

RadialRadial

canonical transforms

Hilbert space  on  R+= [0, ∞)

for m = 0, 1, 2, …

involves Bessel functions

M. Moshinsky, T.H. Seligman and KBW, Canonical radial transformations andthe radial oscillator and Coulomb problems, J. Math. Phys. 13, 901-907 (1972)

The complex extensions include Barut‐Girardello‐type integral transforms

KBW, Canonical transforms II. Complex radial transforms, J. Math. Phys. 15, 2101-2111 (1974)

HyperbolicHyperbolic

canonical transforms

Hilbert space  on  R+ ⊕

R+

for s e [0,∞), ϖ

e

±

involves Hankel

functions:

KBW, Canonical Transforms IV.Hyperbolic transforms:continuous series of SL(2,R)representations,J. Math. Phys. 21, 680-688 (1980)

no complex extension…

What areare

canonical transforms?Valentin Bargmann (1947),  and Gel’fand and Naimark (1947 also)studied the unirreps of the 2+1 Lorentz algebra and group SO(2,1).

We have a 2nd‐order diff op realizationthat transforms under Sp(2,R) as:

The centrifugal/centripetal potential divides unirreps into discretecontinuous( and exceptional){

…and Sp(2,R) has 3 subgroup orbits

hyperbolicparabolic

elliptic(non‐exponential)

These providethe  ‘row’

indicesfor matrix orintegral kernelrepresentations.

Elliptic orbit:lower‐bound orinfinite matrices

Parabolic orbitParabolic orbitintegral kernels oflinear canonical transforms

Hyperbolic orbitintegral kernels

Thus we have2

representation series in

3

subgroup reductions

Hilbert spaces undergoing unitary irreducibleHilbert spaces undergoing unitary irreducibleSp(2,R) Linear Canonical Transforms:Sp(2,R) Linear Canonical Transforms:Representationseries:

ellipticbasis

parabolicbasis

hyperbolicbasis

infinite or half‐infinitediscrete (vector) functions

functions of a radiusor pairs of the same.

functions on the real lineor pairs of the same.

‐elliptic Acts on lower‐bound infinite vectors

‐parabolic Acts on functions on a radius  

LCTs on functions on the real line (Collins, Moshinsky & Quesne)

‐hyperbolic Acts on functions on the real line

‐elliptic Acts on infinite vectors

‐parabolic Acts on two‐functions on a radius  

‐hyperbolic Acts on two‐functions on the real line

Thus

Sp(2,R) provides 6 faces of LCTs at least

In the parabolic basis,In the parabolic basis,The C / M‐Q  LCTs  belong to D(+,1/4) + D(+,3/4) .Radial LCTs belong to the discrete series D(+,k) for k > 0 .Hyperbolic LCTs belong to continuous series C(e,s) for k = 1/2 + i s .In the elliptic basiswe can have lower‐bound infinite matrix LCTs in the discrete series,or infinite matrix LCTs in the continuous series.

The hyperbolic basis for 2‐component functions remains unused.The exceptional interval  0 < k < 1 has 1‐param self‐adjoint extensions.

The exceptional representation series is terra incognita (hic vivunt leones!).

Conserveduncertaintyrelations for `classical’

LCTs

There has been much recent interest

in FINITE LCTs to permit procesing by computers

But of course…

non‐compact groups do not have 

unitary finite‐dimensional faithful representations !!!

Preferably using the FFT algorithm for

So, people use:

A ‘finite fractional Fourier transform’

presents multi‐solutions,and a toroidal phase space will not rotate continuously.

Finite data sets divide us into two cities:in one, those who believe in symmetries,in the other, those who want fast results.

Use discretized

version of LCTsnot

unitary,  do not composetoroidal

phase space a mess…

Use the group SU(2).Enjoy unitarity, composition,and a nice

spherical phase space.

¡ Muchas gracias!

PAPIITPAPIIT--UNAMUNAMand SEPSEP--CONACYTCONACYT

projectsÓptica Matemática

¡ Muchas gracias!

PAPIITPAPIIT--UNAMUNAMand SEPSEP--CONACYTCONACYT

projectsÓptica Matemática

With

best admiration

and

wishesfor

Prof. Francesco Iachello

on

his

70 years

of

success

!

top related